Soft Smart Interfaces for Bilateral Teleoperation with Distributed Tactile Haptic Feedback in Remote Medical Palpation

Abstract

Robotics and teleoperated robotics have been increasingly implemented in the healthcare system in the last two decades. From robotic-assisted minimally invasive surgery to neurosurgery, endoscopy, and remote monitoring of patients at risk, robotic systems have showcased a beyond-human performance, allowing medical procedures to be performed at a distance, in hard-to-reach environments, and with strongly increased accuracy and speed. However, areas like primary care are still to be permeated by the ongoing growth of robotics. In particular, the routine test of physical examination, and especially the task of palpation, cannot be performed remotely, despite several research groups having attempted to develop teleoperated systems to enable remote palpation. This thesis proposes soft smart interfaces for distributed tactile haptic feedback, trying to exploit recent advancements in soft robotics to develop interfaces able to mimic the feeling of hand-to-skin palpation while providing the necessary information to enable the user to rapidly perform the correct diagnosis during remote palpation. Moreover, the focus is centered on both of the human-machine interfaces: the one between doctor and robot and the one between robot and patient. For the sake of safety, the patient is simulated using silicone phantoms of different shapes. Part I is centered around the development of a custom-made teleoperated architecture and the integration of shared control as a tool to augment the user experience while performing remote palpation. A combination of shared control and predictive models can be used to avoid indirect instabilities, normally created by the human operator's overcompensation of the delayed tactile recorded signal. Part II concerns the doctor-robot interface. Such an interface should act as a bilateral physical twin of the patient in the remote environment. It should not only reproduce the morphology of the tissues in contact with the robotic end-effector but also allow the doctor to control the robot's trajectory by just palpating the interface itself. This is achieved thanks to the embodied intelligence of the implemented device due to the smart position of sensors able to exploit the non-linearities introduced by the soft silicones used during manufacturing. All elements of such an interface are tested individually in unilateral teleoperation and are then integrated to achieve a fully functioning bilateral architecture. Part III focuses on the robot-patient interface. In order to ensure contact with the environment through a soft material and avoid mechanical constraints imposed by the sensing technology, this thesis exploits the approach of filtered tactile sensing. However, the presence of a mechanical filter between the sensing unit and the remote environment affects the quality of the recorded data. Hereby, I analyze the requirements of the selected filter in relation to the environment, as the guidelines to achieve the ideal filter cannot be defined without taking into consideration the surroundings and the contact interaction. Moreover, in order to bypass the trade-off between compliance and mechanical filtering, magnetorheological materials have been used to implement online tunable stiffness filters. Furthermore, a self-sensing 3D printable magnetorheological elastomer has been developed, paving the way for closed-loop stiffness control. Lastly, Part IV goes beyond haptics and investigates the importance of secondary modalities. In particular, facial expressions are proven to deliver important information for the diagnosis. Moreover, they showcase good performance without any other feedback modality. In summary, the work in this thesis focuses on the introduction of soft interfaces for remote palpation. By utilizing soft materials it is possible to obtain interfaces that feel like soft tissues while achieving the functionality needed by the teleoperated system. Thanks to a bilateral physical twin as a doctor-robot interface and filtered tactile sensing on the robot-patient interface, this thesis showcases how to record and deliver distributed tactile haptic feedback, thus paving the way for the development of teleoperated remote palpation systems.